Showing papers by "Roma Tre University published in 1993"

Spectral electromagnetic modeling of a planar integrated structure with a general grounded anisotropic slab

Abstract: A formulation for the solution of the electromagnetic field in vertically stratified structures with a general anisotropic grounded slab under planar electric excitation conditions is presented. The development of the solution is obtained by using the two-dimensional Fourier transform applied to the partial differential equations describing the electromagnetic field, which can be expressed in terms of dyadic electric Green's functions. Moreover, a spectral equivalent circuit representation of the solution, separated into its spectral transverse electric (TE) and transverse magnetic (TM) components (with respect to the vertical direction), is given. It is demonstrated that such a spectral circuit representation is always possible when the constitutive epsilon and mu tensors are reduced to their diagonal form. The transmission-line representation for an unbounded gyrotropic medium with vertical preferred axis is presented. This theory is applied to a planar structure with an anisotropic biaxial grounded slab and an electric horizontal point source embedded in it. Numerical examples of evaluation of the radiated pattern and noteworthy information about the radiation on the horizon plane are given. >

Very efficient VLSI implementation of CNN with discrete templates

Abstract: Hardware implementation of programmable neural networks requires multiplications of input analogue voltages and constant values. In the Letter a very efficient implementation of cellular neural networks with digital templates is presented.

MODA: moving object detecting architecture

Abstract: A type of cellular neural network (CNN) is described, which may be classified in the broader category of generalized cellular neural networks (GCNNs). Its novelty consists both in the task it performs and in its architecture and way of operation. The input to the network is a two-dimensional picture that is processed continuously in order to detect real time trajectories of moving objects in a noisy environment. MODA is designed by synthesis, so that it does not require learning, and it performs its task by implementing a nonlinear continuous functional in a vector space. The network, its architecture, its equations, and the method of design are described. In addition, the new network is compared with known paradigms of ANN and CNN. Results of simulations are also reported. >

VLSI implementation of a Cellular Neural Network with programmable control operator

Abstract: Cellular Neural Networks (CNN) are a particular class of neural networks based on a regular structure. Using this property a suitable architecture can be designed for a very efficient analog implementation. The core of a cellular neuron is an analog multiplier that can be implemented by using different approaches. In particular, if the CNN is used for conventional applications, as for example Connected Component Detector (CCD), different solutions are possible in terms of fixed or programmable cloning template. In this paper a VLSI implementation of programmable CNN with control operator B and symmetrical and anti symmetrical feedback operator A is presented. >

Power distributed amplifier based on interdigital combiners

Abstract: An innovative power distributed amplifier topology to improve the small signal gain and the power added efficiency is presented. The introduction of Lange couplers at the input and output ports of the amplifier leads to a remarkable advantage in comparison with the conventional configuration. An example is discussed to confirm the validity and the effectiveness of the proposed circuit.

Characterization of the resonant and coupling parameters of dielectric resonators of NRD-guide filtering devices

Abstract: The design of millimeter-wave filtering devices using dielectric resonators of various shapes in nonradiative dielectric (NRD) waveguides is considered. By developing different approximate and rigorous theoretical methods, the resonant and coupling properties (frequencies, field configuration, unloaded and loaded quality factors, scattering parameters, etc.) are modeled and implemented in an adaptive code. Experimental comparison through suitable measurements are performed to validate the theoretical and numerical results. >

Temporal properties of the timed d-graphs of distributed processes

Abstract: Some concepts and tools used to verify the time boundaries of sequential software for hard-real-time applications are based on program control flow and the execution time of instructions or blocks. The paper extends a structural approach from sequential to distributed software. It is conceptually and practically based on timed Concurrent Graphs, TCGs, which extend the concept of sequential program d-graphs. The properties of the concurrent exploration of a 'system' of timed concurrent d-graphs are considered, and TCGs are shown to be useful tools for predicting the initial and the steady state worst temporal behaviours of life-time critical distributed static or dynamic systems of messaging only processes, in the maximum pure parallel model. The semantics of Ada-like tasking and communication models are considered. Also, obstacles to predictability are identified and predictable solutions are provided. The proposed approach holds for different tasking and communication models, as shown by an included CSP-like case-study. >